NFS causing headache in CoreOS Docker Containers

carstend

Hi all,

I hope this is the right forum to ask this question ...

I've got an issue with NFS stale file handles in some containers, which I cannot resolve.

Setup:
NAS, provides SMB and NFS shares
2 PCs with Windows 10, running VirtualBox with several CoreOS VMs with a number of containers each
2nd NAS for Backup of main NAS

Since CoreOS does not natively support SMB, I use the NAS NFS share for:
configuration of the CoreOS Instances (ignition files)
the containers (yaml files)
container data

The reason for this configuration is that I've got everything centrally stored in one place, which I can manage across the instances. Also, some containers on different VMs/hosts can share the same data.

Unfortunately, after some traffic on the NAS some containers stop working properly, since the NFS files handle becomes stale. I've been through a lot of research of tons of forums and I found loads of ways how to re-mount the share, but re-mounting is a work-around and not a real cure of the problem.

Now, since I can't change the behaviour of NFS and CoreOS does not natively support SMB, what else could one use as a "shared" structure across the VMs and the hosts?
How do VM farms share their data? I would really like to keep the NAS as my central data storage for the CoreOS Instances, Containers and their data.

Do I miss the obvious?

Regards

Carsten